Improved quantification of MRI relaxation rates using Bayesian estimation

Traditional magnetic resonance imaging (MRI) studies are based on image contrast and qualitative analysis. However, there is an increasing interest in quantifying the physical parameters of the object such as the free induction decay rate, T*₂ . In this paper, a new Bayesian algorithm is proposed for the estimation of T*₂ from gradient echo MRI scans. Current estimation methods use a simple signal model based on Fourier reconstruction which imposes a trade-off between the signal-to-noise ratio (SNR) and image distortion, and results in estimation bias. The proposed algorithm uses a Gibbs sampler in a Bayesian framework to account for image distortion allowing data samples to be acquired with increased SNR, improving the estimation accuracy. Estimation results on simulated objects and in vivo experimental data demonstrate the effectiveness of the algorithm.